Re: Definition of "External" Host

[Eric Brunner-Williams in Portland Maine <brunner@nic-naa.net>]

Scott,

It appears to me that there is some confusion concerning the following:

	1. the existance distinct policies scoped to the presence
	   (or absence) of infix dots, viz:
	  	*.a.z	policy "az",
	   	*.b.z	policy "bz",
	   	*.z	policy " z",

	2. the delegation of authoritative publication requirements,
	    viz:
	    	existence of glue records for nameservers (aka "hosts"),

	3. distinct instances of EPP servers.

In the context of the channel capacity of a single BEEP session I discussed
the possibility of a single TCP connection servicing the command/response
exchanges of two EPP endpoints conducting transactions against distinct
registries. This was in an exchange with Sheer El-Showk, back in August of
'01. That is what caught my limited attention in this exchange -- operators
with multiple policies and possibly using some mechanism other than distinct
endpoints to distinguish between policy scopes.

Comments:

New.Net marketed infix dots without delegation, and some other operators
may similarly market infix w/o delegation. Utility arguements for infix
dots w/o delegation are outside of our scope. This is fortunate for me,
as I can't think of any off-hand.

The text in -5 (host) at 2.5 isn't quite consistent with the text at 3.2,
the first has "SHOULD be required only as needed", the second a "REQUIRED
only as needed". However, we missed the point of using stub zones to
obviate the need for glue NS records in a parent zone.

I'd appreciate it if the discussion were less hypothetical, and some of
the discussion conducted in master file format.

Eric